Gemini 3.6 Flash
Gemini 3.6 Flash is Google's generally available Flash workhorse model for coding, multimodal analysis, knowledge work, and multi-step agents. Its Gemini API model ID is gemini-3.6-flash.
Gemini 3.6 Flash improves the quality and token efficiency of the previous Flash tier while reducing output-token pricing. It is the practical migration target for teams using Gemini 3.5 Flash or older Gemini Flash models in production agents.
Google announced Gemini 3.6 Flash on July 21, 2026 and documents it as GA in the Gemini API. The model accepts text, image, audio, and video input, returns text, supports a 1M-token context window and 64K maximum output, and defaults to medium thinking. Google lists standard pricing at $1.50 per million input tokens and $7.50 per million output tokens. Google-reported evaluations show 17% fewer output tokens than 3.5 Flash on the Artificial Analysis Index and gains in coding, computer use, ML research, and knowledge work; these launch results should still be validated on each production workload.
- Run multi-step coding and repository agents with built-in tools.
- Analyze documents, charts, images, audio, and video with long multimodal context.
- Automate knowledge-work workflows that need stronger quality than earlier Flash models.
- Migrate Gemini 3.5 Flash workloads to a lower output-token price.
Gemini 3.6 Flash is GA under the model ID gemini-3.6-flash. Google documents a 1M-token context window, 64K maximum output, text/image/audio/video input, text output, thinking, and built-in tools including Computer Use. It is available through the Gemini API, Google AI Studio, Gemini Enterprise, the Gemini app, and Google Antigravity.
Google positions 3.6 Flash as better at coding, knowledge work, multimodal tasks, and agentic execution while using fewer reasoning steps, tool calls, and output tokens. Standard input pricing remains $1.50 per million tokens, while output pricing drops from 3.5 Flash's $9.00 to $7.50 per million tokens.
- Google reports 17% fewer output tokens than 3.5 Flash on the Artificial Analysis Index.
- Google reports DeepSWE at 49% versus 37% and OSWorld-Verified at 83.0% versus 78.4%.
- Treat launch benchmarks as vendor-reported screening evidence, not proof for a specific repository or agent harness.
The Gemini API migration guide says new models ignore and will later reject temperature, top_p, and top_k, no longer allow prefilled model turns, and use thinking_level instead of thinking_budget. Migrations should verify structured output, function calls, multimodal tool responses, latency, token use, and task success before replacing a production model.
Gemini 3.6 Flash FAQ
Common questions about Gemini 3.6 Flash.
What is the Gemini 3.6 Flash API model ID?+
The generally available Gemini API model ID is gemini-3.6-flash.
Is Gemini 3.6 Flash cheaper than Gemini 3.5 Flash?+
Yes for standard output tokens. Both models cost $1.50 per million input tokens, while Gemini 3.6 Flash costs $7.50 per million output tokens versus $9.00 for 3.5 Flash. Google also reports lower output-token usage on its evaluation set, but total task cost depends on your prompts, tools, retries, and success rate.
What are the Gemini 3.6 Flash context and output limits?+
Google documents a 1M-token context window and a 64K maximum output for Gemini 3.6 Flash.
Does Gemini 3.6 Flash support computer use?+
Yes. Google documents Computer Use as a built-in client-side tool for Gemini 3.6 Flash through supported Gemini API and enterprise surfaces. Applications still need their own permission, safety, and action-confirmation controls.